Risk management is the single skill that separates traders who survive long-term from those who blow up their accounts in the first bear market. No edge, no strategy, no indicator saves you without it. Most traders spend 95% of their time searching for winning trades and 5% thinking about what happens when they lose — it should be the reverse.
Here's the counterintuitive truth: a trader who wins 40% of trades but has impeccable risk management will significantly outperform a trader who wins 70% of trades but sizes recklessly. The reason is the asymmetry of drawdown recovery.
| Loss suffered | Gain required to fully recover | Recovery time at 10%/month |
|---|---|---|
| 10% | 11.1% | ~1 month |
| 25% | 33.3% | ~3 months |
| 50% | 100% | ~8 months |
| 75% | 300% | ~2.5 years |
| 90% | 900% | ~7+ years (unlikely) |
This table explains why protecting your capital is more valuable than maximising returns. The time cost of recovering from a large drawdown is so severe that most traders who blow up 50%+ simply never recover — they either run out of patience or make increasingly desperate decisions trying to get back to even.
Never risk more than 1% of your total account on a single trade. On a $10,000 account, that's $100 maximum loss per position. This feels uncomfortably small — good. That discomfort is the psychological resistance that separates disciplined traders from gamblers.
Why 1% specifically? It means 10 consecutive full losses only costs 10% of your account. At that scale, your edge has ample time to manifest before you're in trouble. And 10 consecutive full losses is extreme even for poor strategies.
For prediction markets: 1–2% is appropriate. Polymarket positions have defined maximum loss (your stake) and defined maximum gain (contract payout), making the 1% rule easier to apply than stop-loss-based crypto trading.
Position size is not "how much I feel like buying." It's a calculation based on your bankroll, your risk per trade, and the distance to your stop-loss (or maximum acceptable loss).
Example — Crypto trade: $10,000 account, 1% risk ($100). Buy BTC at $95,000, stop at $93,100 ($1,900 risk/BTC). Position size = $100 ÷ $1,900 = 0.053 BTC ($5,000 notional). Note: your risk is only $100 even though the notional is $5,000.
Example — Polymarket: $5,000 bankroll, 2% risk ($100). Market priced at 60¢. Stake = $100 ÷ (1 − 0.60) = $250 maximum position (risking $100 if it resolves NO, winning $167 if YES).
A stop-loss is your pre-determined exit price — set before you enter a trade. The critical rule: set it based on technical/fundamental logic, not based on how much loss you're willing to accept emotionally.
Mental stop-losses don't count. When a trade is moving against you, the human brain produces a cascade of rationalisations for why this time it will turn around. Hard stops (automated orders) execute without emotional override.
In prediction markets, "stop-losses" work differently — markets don't move continuously, so you apply the principle differently:
Diversification in crypto requires understanding correlation — not just holding many assets. During crypto market crashes, BTC, ETH, altcoins, and many crypto-adjacent prediction markets all fall simultaneously. Holding 10 crypto assets is not diversification — it's concentrated crypto exposure with extra steps.
True diversification requires positions that move independently:
The Kelly Criterion is the formula that calculates the mathematically optimal fraction of your bankroll to bet on each trade to maximise long-run growth:
Where: b = net odds (payout ratio), p = probability of winning, q = probability of losing (1 − p).
Example: Polymarket market at 40¢ that you estimate has a 55% true probability. Payout = 1/0.40 − 1 = 1.5× your stake. Kelly = (1.5 × 0.55 − 0.45) ÷ 1.5 = 23% of bankroll. Full Kelly says bet 23%.
Why fractional Kelly: Full Kelly is mathematically optimal but produces terrifying volatility — drawdowns of 50%+ are common. Most professionals use 25–50% of Kelly (fractional Kelly): sacrificing ~10% of expected growth rate in exchange for 75% less variance. For a $5,000 bankroll, fractional Kelly on the above example = 5.75–11.5% = $290–$575.
Crypto trading risk management frameworks don't map perfectly onto prediction markets. Polymarket has structural features that require adapted rules — and unique risks that standard trading guides completely ignore.
Polymarket hosts dozens of active crypto price markets simultaneously: "Will BTC exceed $100k?", "Will ETH exceed $5k?", "Will crypto total market cap exceed $5T?". These markets are highly correlated — they'll all swing in the same direction when a macro crypto event hits. A trader who's 2% into each of 15 correlated crypto markets effectively has a 30% concentrated position in crypto sentiment. One bad news day destroys them.
Solution: Count your thematic exposure, not just your per-market sizing. Cap total crypto sentiment exposure at 10–15% of bankroll regardless of how many individual positions make up that exposure.
Unlike crypto spot trading where price moves are continuous, prediction market outcomes are binary and final. You can have a genuinely correct probability estimate (55% → you bet YES) and still lose 45% of the time by definition. This variance means short-run results are meaningless as performance indicators — only run hundreds of markets long is your win rate statistically significant.
Implication: Never increase bet size after losses to "recover" (Martingale fallacy). Never decrease after wins thinking you're "due for a loss." Size based on your bankroll at that moment and your edge estimate for that specific market — not your recent run.
Many Polymarket markets have wide bid-ask spreads, particularly in lower-volume or newly listed markets. Entering a 60¢ market with a 4% spread means your effective entry is 62¢ — 3.3% immediate slippage. This dramatically reduces the edge on marginal trades. Apply a minimum edge threshold: only enter a market when your estimated probability exceeds the market price by at least the spread plus a cushion (typically 5–8% for thin markets).
Keep 10–20% of your prediction market bankroll as uninvested USDC at all times. This serves two functions: (1) It provides capital to add to high-conviction positions when markets move against you on valid theses — the best time to buy is often when the market disagrees with you most sharply. (2) It prevents the psychological pressure of being fully deployed — "I have no USDC left, I can't act on this obvious mispricing." Opportunity cost of idle capital is real, but the cost of being unable to act at critical moments is higher.
Before entering any Polymarket position, verify:
A 60% win rate on Polymarket looks great — but if your average win is $30 and your average loss is $80, you're losing money with a majority win rate. Track these four metrics monthly:
The 1% rule states you should never risk more than 1% of your total trading capital on any single trade. On a $10,000 bankroll, that's $100 maximum loss per trade. This ensures even a 10-trade losing streak costs only 10% — survivable. It applies equally to prediction markets: never risk more than 1–2% of bankroll on any single Polymarket position.
Kelly Criterion is a formula calculating the optimal fraction of bankroll to bet: f* = (b×p − q) ÷ b (where b = net odds, p = win probability, q = loss probability). Full Kelly maximises long-run growth but produces extreme volatility. Use fractional Kelly at 25–50% — you sacrifice ~10% of expected growth rate in exchange for dramatically lower variance. Use our Kelly Calculator to compute it instantly.
Drawdown is the decline from your portfolio peak to its current trough. A 50% drawdown requires a 100% gain to recover. Manage drawdown by: (1) Never risking more than 1–2% per trade. (2) Reducing position size by 50% once you've drawn down 20% from peak. (3) Stopping trading entirely if drawdown exceeds 35% — something is wrong with your strategy, not just your luck.
Key differences: (1) Outcomes are binary and final — no stop-losses mid-market. (2) Correlation risk is hidden — many Polymarket markets (especially crypto) move together. (3) Liquidity risk is higher in thin markets with wide spreads. (4) Resolution risk is unique — even correct probability estimates lose nearly half the time by mathematical necessity. Apply thematic exposure caps and minimum edge thresholds in addition to per-position sizing rules.